Complete the changes for supporting CSS.

asr/python/gen_asrinput_raw_continuous.py

@@ -103,17 +103,22 @@ def main(args):
     # Create an ASR script. 
     with open(os.path.join(decoding_cmd, 'decode.sh'),'w') as f:
         cmd = 'sh {} {} {} . {}'.format(os.path.join(tool_path, 'run_asr_continuous.sh'), 
-                                   os.path.join(decoding_cmd, 'meeting_list.scp'), 
-                                   decoding_result, 
-                                   am_path)
+                                        os.path.join(decoding_cmd, 'meeting_list.scp'), 
+                                        decoding_result, 
+                                        am_path)
         f.write(cmd+'\n')
-        cmd = 'python {} --inputdir {} --outputdir {}'.format(os.path.normpath(os.path.join(tool_path, '../python/sortctm.py')), 
-                                                              decoding_result, 
-                                                              decoding_result + '.sorted')
+        if args.multi_stream:
+            cmd = 'python {} --with_channel --inputdir {} --outputdir {}'.format(os.path.normpath(os.path.join(tool_path, '../python/sortctm.py')), 
+                                                                decoding_result, 
+                                                                decoding_result + '.sorted')
+        else:
+            cmd = 'python {} --inputdir {} --outputdir {}'.format(os.path.normpath(os.path.join(tool_path, '../python/sortctm.py')), 
+                                                                decoding_result, 
+                                                                decoding_result + '.sorted')
         f.write(cmd+'\n')
         cmd = 'chown -R {}:{} {}'.format(os.getuid(), os.getgid(), decoding_result) 
         f.write(cmd+'\n')
-        cmd = 'chown -R {}:{} {}'.format(os.getuid(), os.getgid(), decoding_result + '.sort') 
+        cmd = 'chown -R {}:{} {}'.format(os.getuid(), os.getgid(), decoding_result + '.sorted') 
         f.write(cmd+'\n')
 
 
@@ -134,6 +139,9 @@ def make_argparse():
     parser.add_argument('--merge_margin', default=1, metavar='<float>', type=float, 
                         help='Segmentation parameter.')
 
+    parser.add_argument('--multi_stream', action='store_true', 
+                        help='Set this flag when processing CSS (or multi-stream) outputs.')
+
     return parser
 
 

asr/python/sortctm.py

@@ -4,15 +4,19 @@
 from collections import defaultdict
 
 
-def sortout_results(ctmfile, tgtdir):
+def sortout_results(ctmfile, tgtdir, with_channel=False):
     print('')
     print('{}'.format(ctmfile))
     print('    -> {}'.format(tgtdir))
 
     os.makedirs(tgtdir, exist_ok=True)
 
-    # overlap_ratio_0.0_sil2.9_3.0_session0_actual0.0_segment_5_195199_384160 1 1.770 0.420 REMEMBER
-    ptrn = re.compile('^overlap_ratio_([\d\.]+)_sil([\d\.]+)_([\d\.]+)_session(\d+)_actual([\d\.]+)_segment_(\d+)_(\d+)_(\d+)$')
+    if with_channel: 
+        # overlap_ratio_0.0_sil0.1_0.5_session0_actual0.0_segment_6_1_399200_556960 1 7.330 0.550 SPEAKING
+        ptrn = re.compile('^overlap_ratio_([\d\.]+)_sil([\d\.]+)_([\d\.]+)_session(\d+)_actual([\d\.]+)_[A-Za-z]+_(\d+)_\d+_(\d+)_(\d+)$')
+    else:
+        # overlap_ratio_0.0_sil2.9_3.0_session0_actual0.0_segment_5_195199_384160 1 1.770 0.420 REMEMBER
+        ptrn = re.compile('^overlap_ratio_([\d\.]+)_sil([\d\.]+)_([\d\.]+)_session(\d+)_actual([\d\.]+)_[A-Za-z]+_(\d+)_(\d+)_(\d+)$')
 
     results = defaultdict(list)
 
@@ -72,24 +76,7 @@ def main(args):
             cfg = os.path.splitext(os.path.basename(ctmfile))[0]
             tgtdir = os.path.join(args.outputdir, cfg, cond)
 
-            sortout_results(ctmfile, tgtdir)
-
-
-
-
-    # # Write the sorted version to the output CTM file. 
-    # for segment in hyps:
-    #     outputdir = os.path.join(args.outputdir, os.path.splitext(os.path.basename(args.inputfile))[0])
-    #     outputfile = os.path.join(outputdir, segment + '.ctm')
-
-    #     os.makedirs(outputdir, exist_ok=True)
-
-    #     curhyps = sorted(hyps[segment], key=lambda hyp: hyp[2])
-
-    #     with open(outputfile, 'w') as f:
-    #         for mtg, ch, start, dur, wrd, conf in curhyps:
-    #             print('{} {} {:.2f} {} {} {}'.format(mtg, ch, start, dur, wrd, conf), file=f)
-
+            sortout_results(ctmfile, tgtdir, with_channel=args.with_channel)
 
 
 
@@ -101,6 +88,8 @@ def make_argparse():
                         help='Raw ASR result directory.')
     parser.add_argument('--outputdir', metavar='<path>', required=True, 
                         help='Output directory.')
+    parser.add_argument('--with_channel', action='store_true', 
+                        help='Has a channel index in each CTM line.')
 
     return parser
 

asr/scripts/gen_asrinput_raw_continuous.sh

@@ -2,12 +2,13 @@
 
 PYTHON=python
 
+curdir=`dirname $0`
+curdir=`realpath $curdir`
+
 # args
-prep_script=../python/gen_asrinput_raw_continuous.py
+prep_script=$curdir/../python/gen_asrinput_raw_continuous.py
 libricss_path=$EXPROOT/data
-script_path=../scripts
+script_path=$curdir/../scripts
 asr_path=$AMPATH
 
-echo "$prep_script --data_path $libricss_path --tool_path $script_path --asr_path $asr_path"
-
-$PYTHON $prep_script --data_path $libricss_path --tool_path $script_path --asr_path $asr_path
+$PYTHON $prep_script --input_path $libricss_path/monaural/segments --decode_path $libricss_path/baseline/segments --tool_path $script_path --am_path $asr_path

asr/scripts/gen_asrinput_separated_continuous.sh

@@ -2,13 +2,15 @@
 
 PYTHON=python
 
+curdir=`dirname $0`
+curdir=`realpath $curdir`
+
 # args
-prep_script=../python/gen_asrinput_separated_continuous.py
+prep_script=$curdir/../python/gen_asrinput_raw_continuous.py
 libricss_path=$EXPROOT/data
-script_path=../scripts
+script_path=$curdir/../scripts
 asr_path=$AMPATH
 
-echo "$prep_script --data_path $libricss_path --tool_path $script_path --asr_path $asr_path"
+$PYTHON $prep_script --multi_stream --input_path $libricss_path/separation_result/continuous_separation --decode_path $libricss_path/separation_baseline --tool_path $script_path --am_path $asr_path
 
-$PYTHON $prep_script --data_path $libricss_path --tool_path $script_path --asr_path $asr_path